distancecommand.cpp

   1 /*
   2  *  distancecommand.cpp
   3  *  Mothur
   4  *
   5  *  Created by Sarah Westcott on 5/7/09.
   6  *  Copyright 2009 Schloss Lab UMASS Amherst. All rights reserved.
   7  *
   8  */
   9
  10 #include "distancecommand.h"
  11 #include "ignoregaps.h"
  12 #include "eachgapdist.h"
  13 #include "eachgapignore.h"
  14 #include "onegapdist.h"
  15 #include "onegapignore.h"
  16
  17 //**********************************************************************************************************************
  18
  19 DistanceCommand::DistanceCommand(string option){
  20         try {
  21                 abort = false;
  22                 Estimators.clear();
  23
  24                 //allow user to run help
  25                 if(option == "help") { help(); abort = true; }
  26
  27                 else {
  28                         //valid paramters for this command
  29                         string Array[] =  {"fasta", "phylip", "calc", "countends", "cutoff", "processors"};
  30                         vector<string> myArray (Array, Array+(sizeof(Array)/sizeof(string)));
  31
  32                         OptionParser parser(option);
  33                         map<string, string> parameters = parser.getParameters();
  34
  35                         ValidParameters validParameter;
  36
  37                         //check to make sure all parameters are valid for command
  38                         for (map<string, string>::iterator it2 = parameters.begin(); it2 != parameters.end(); it2++) {
  39                                 if (validParameter.isValidParameter(it2->first, myArray, it2->second) != true) {  abort = true;  }
  40                         }
  41
  42                         //check for required parameters
  43                         fastafile = validParameter.validFile(parameters, "fasta", true);
  44                         if (fastafile == "not found") { mothurOut("fasta is a required parameter for the dist.seqs command."); mothurOutEndLine(); abort = true; }
  45                         else if (fastafile == "not open") { abort = true; }
  46                         else{
  47                                 ifstream inFASTA;
  48                                 openInputFile(fastafile, inFASTA);
  49                                 alignDB = SequenceDB(inFASTA);
  50                                 inFASTA.close();
  51                         }
  52
  53                         //check for optional parameter and set defaults
  54                         // ...at some point should added some additional type checking...
  55                         calc = validParameter.validFile(parameters, "calc", false);
  56                         if (calc == "not found") { calc = "onegap";  }
  57                         else {
  58                                  if (calc == "default")  {  calc = "onegap";  }
  59                         }
  60                         splitAtDash(calc, Estimators);
  61
  62                         string temp;
  63                         temp = validParameter.validFile(parameters, "countends", false);        if(temp == "not found"){        temp = "T";     }
  64                         convert(temp, countends);
  65
  66                         temp = validParameter.validFile(parameters, "cutoff", false);           if(temp == "not found"){        temp = "1.0"; }
  67                         convert(temp, cutoff);
  68
  69                         temp = validParameter.validFile(parameters, "processors", false);       if(temp == "not found"){        temp = "1"; }
  70                         convert(temp, processors);
  71
  72                         phylip = validParameter.validFile(parameters, "phylip", false);         if(phylip == "not found"){      phylip = "F"; }
  73
  74
  75                         ValidCalculators validCalculator;
  76
  77                         if (isTrue(countends) == true) {
  78                                 for (int i=0; i<Estimators.size(); i++) {
  79                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
  80                                                 if (Estimators[i] == "nogaps")                  {       distCalculator = new ignoreGaps();      }
  81                                                 else if (Estimators[i] == "eachgap")    {       distCalculator = new eachGapDist();     }
  82                                                 else if (Estimators[i] == "onegap")             {       distCalculator = new oneGapDist();      }
  83                                         }
  84                                 }
  85                         }else {
  86                                 for (int i=0; i<Estimators.size(); i++) {
  87                                         if (validCalculator.isValidCalculator("distance", Estimators[i]) == true) {
  88                                                 if (Estimators[i] == "nogaps")          {       distCalculator = new ignoreGaps();                                      }
  89                                                 else if (Estimators[i] == "eachgap"){   distCalculator = new eachGapIgnoreTermGapDist();        }
  90                                                 else if (Estimators[i] == "onegap")     {       distCalculator = new oneGapIgnoreTermGapDist();         }
  91                                         }
  92                                 }
  93                         }
  94
  95                 }
  96
  97         }
  98         catch(exception& e) {
  99                 errorOut(e, "DistanceCommand", "DistanceCommand");
 100                 exit(1);
 101         }
 102 }
 103
 104 //**********************************************************************************************************************
 105
 106 DistanceCommand::~DistanceCommand(){
 107
 108         for(int i=0;i<lines.size();i++){
 109                 delete lines[i];
 110         }
 111
 112 }
 113
 114 //**********************************************************************************************************************
 115
 116 void DistanceCommand::help(){
 117         try {
 118                 mothurOut("The dist.seqs command reads a file containing sequences and creates a distance file.\n");
 119                 mothurOut("The dist.seqs command parameters are fasta, calc, countends, cutoff and processors.  \n");
 120                 mothurOut("The fasta parameter is required.\n");
 121                 mothurOut("The calc parameter allows you to specify the method of calculating the distances.  Your options are: nogaps, onegap or eachgap. The default is onegap.\n");
 122                 mothurOut("The countends parameter allows you to specify whether to include terminal gaps in distance.  Your options are: T or F. The default is T.\n");
 123                 mothurOut("The cutoff parameter allows you to specify maximum distance to keep. The default is 1.0.\n");
 124                 mothurOut("The processors parameter allows you to specify number of processors to use.  The default is 1.\n");
 125                 mothurOut("The dist.seqs command should be in the following format: \n");
 126                 mothurOut("dist.seqs(fasta=yourFastaFile, calc=yourCalc, countends=yourEnds, cutoff= yourCutOff, processors=yourProcessors) \n");
 127                 mothurOut("Example dist.seqs(fasta=amazon.fasta, calc=eachgap, countends=F, cutoff= 2.0, processors=3).\n");
 128                 mothurOut("Note: No spaces between parameter labels (i.e. calc), '=' and parameters (i.e.yourCalc).\n\n");
 129         }
 130         catch(exception& e) {
 131                 errorOut(e, "DistanceCommand", "help");
 132                 exit(1);
 133         }
 134 }
 135 //**********************************************************************************************************************
 136
 137 int DistanceCommand::execute(){
 138         try {
 139
 140                 if (abort == true) { return 0; }
 141
 142                 int numSeqs = alignDB.getNumSeqs();
 143                 cutoff += 0.005;
 144
 145                 string outputFile;
 146
 147                 //doses the user want the phylip formatted file as well
 148                 if (isTrue(phylip) == true) {
 149                         outputFile = getRootName(fastafile) + "phylip.dist";
 150                         remove(outputFile.c_str());
 151
 152                         //output numSeqs to phylip formatted dist file
 153                 }else { //user wants column format
 154                         outputFile = getRootName(fastafile) + "dist";
 155                         remove(outputFile.c_str());
 156                 }
 157
 158                 //#     if defined (_WIN32)
 159                 //figure out how to implement the fork and wait commands in windows
 160                 //      driver(distCalculator, seqDB, 0, numSeqs, distFile, phylipFile, cutoff);
 161                 //#     endif
 162
 163
 164 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 165                 //if you don't need to fork anything
 166                 if(processors == 1){
 167                         driver(0, numSeqs, outputFile, cutoff);
 168                 }else{ //you have multiple processors
 169
 170                         for (int i = 0; i < processors; i++) {
 171                                 lines.push_back(new linePair());
 172                                 lines[i]->start = int (sqrt(float(i)/float(processors)) * numSeqs);
 173                                 lines[i]->end = int (sqrt(float(i+1)/float(processors)) * numSeqs);
 174                         }
 175
 176                         createProcesses(outputFile);
 177
 178                         map<int, int>::iterator it = processIDS.begin();
 179                         rename((outputFile + toString(it->second) + ".temp").c_str(), outputFile.c_str());
 180                         it++;
 181
 182                         //append and remove temp files
 183                         for (; it != processIDS.end(); it++) {
 184                                 appendFiles((outputFile + toString(it->second) + ".temp"), outputFile);
 185                                 remove((outputFile + toString(it->second) + ".temp").c_str());
 186                         }
 187                 }
 188 #else
 189                 ifstream inFASTA;
 190                 driver(0, numSeqs, outputFile, cutoff);
 191 #endif
 192
 193                 delete distCalculator;
 194
 195                 return 0;
 196
 197         }
 198         catch(exception& e) {
 199                 errorOut(e, "DistanceCommand", "execute");
 200                 exit(1);
 201         }
 202 }
 203 /**************************************************************************************************/
 204 void DistanceCommand::createProcesses(string filename) {
 205         try {
 206 #if defined (__APPLE__) || (__MACH__) || (linux) || (__linux)
 207                 int process = 0;
 208                 processIDS.clear();
 209
 210                 //loop through and create all the processes you want
 211                 while (process != processors) {
 212                         int pid = fork();
 213
 214                         if (pid > 0) {
 215                                 processIDS[lines[process]->end] = pid;  //create map from line number to pid so you can append files in correct order later
 216                                 process++;
 217                         }else if (pid == 0){
 218                                 driver(lines[process]->start, lines[process]->end, filename + toString(getpid()) + ".temp", cutoff);
 219                                 exit(0);
 220                         }else { mothurOut("unable to spawn the necessary processes."); mothurOutEndLine(); exit(0); }
 221                 }
 222
 223                 //force parent to wait until all the processes are done
 224                 for (map<int, int>::iterator it = processIDS.begin(); it != processIDS.end(); it++) {
 225                         int temp = it->second;
 226                         wait(&temp);
 227                 }
 228 #endif
 229         }
 230         catch(exception& e) {
 231                 errorOut(e, "DistanceCommand", "createProcesses");
 232                 exit(1);
 233         }
 234 }
 235
 236 /**************************************************************************************************/
 237 /////// need to fix to work with calcs and sequencedb
 238 int DistanceCommand::driver(int startLine, int endLine, string dFileName, float cutoff){
 239         try {
 240
 241                 int startTime = time(NULL);
 242
 243                 //column file
 244                 ofstream outFile(dFileName.c_str(), ios::trunc);
 245                 outFile.setf(ios::fixed, ios::showpoint);
 246                 outFile << setprecision(4);
 247
 248                 if(isTrue(phylip) && startLine == 0){   outFile << alignDB.getNumSeqs() << endl;        }
 249                 for(int i=startLine;i<endLine;i++){
 250                         if(isTrue(phylip))      {       outFile << alignDB.get(i).getName() << '\t';    }
 251                         for(int j=0;j<i;j++){
 252                                 distCalculator->calcDist(alignDB.get(i), alignDB.get(j));
 253                                 double dist = distCalculator->getDist();
 254
 255                                 if(dist <= cutoff){
 256                                         if (!isTrue(phylip)) { outFile << alignDB.get(i).getName() << ' ' << alignDB.get(j).getName() << ' ' << dist << endl; }
 257                                 }
 258                                 if (isTrue(phylip)) {  outFile << dist << '\t'; }
 259
 260                         }
 261
 262                         if (isTrue(phylip) == true) { outFile << endl; }
 263
 264                         if(i % 100 == 0){
 265                                 mothurOut(toString(i) + "\t" + toString(time(NULL) - startTime)); mothurOutEndLine();
 266                         }
 267
 268                 }
 269                 mothurOut(toString(endLine-1) + "\t" + toString(time(NULL) - startTime)); mothurOutEndLine();
 270
 271                 outFile.close();
 272
 273                 return 1;
 274         }
 275         catch(exception& e) {
 276                 errorOut(e, "DistanceCommand", "driver");
 277                 exit(1);
 278         }
 279 }
 280
 281 /**************************************************************************************************/
 282 void DistanceCommand::appendFiles(string temp, string filename) {
 283         try{
 284                 ofstream output;
 285                 ifstream input;
 286
 287                 //open output file in append mode
 288                 openOutputFileAppend(filename, output);
 289                 openInputFile(temp, input);
 290
 291                 while(char c = input.get()){
 292                         if(input.eof())         {       break;                  }
 293                         else                            {       output << c;    }
 294                 }
 295
 296                 input.close();
 297                 output.close();
 298         }
 299         catch(exception& e) {
 300                 errorOut(e, "DistanceCommand", "appendFiles");
 301                 exit(1);
 302         }
 303 }
 304 /**************************************************************************************************/